Research On Risk Identification And Assessment For Remote Control Maritime Autonomous Surface Ships

The shipping industry has been encountering many challenges,e.g.,shortcoming of seafarers,strict environmental restrictions,frequent occurrence of accidents,etc.To tackle these problems,a research on Maritime Autonomous Surface Ships(MASS)may provide a key.Moreover,the development of MASS can also power the global economic transition in the post-pandemic period.Moreover,as the remote control MASS would be highly seen in the foreseeable future possible,the function of remote control for MASS has been added as a significant part in Guidelines of Smart Ship(2020)stipulated by China Classification Society(CCS).A fundamental issue for shipping is its navigational safety.At the initial phase of research and development on MASS,analyzing,identifying the navigational risk influencing factors(RIFs)of a remote control MASS(without crews onboard)and assessing its navigational risk,is of significance for maintaining safety of the maritime autonomous ecosystem.This thesis focuses on 4 objectives with respect to the navigational risk of a remote control MASS(without crews onboard): 1)propose a model to identify navigational RIFs in different navigation phases;2)develop an approach to compare the risk in different navigational control model and to quantify the uncertainty of such risk;3)design a risk matrix to measure the risk level given probability and consequence;4)construct a Bayesian Belief Network(BBN)model to assess navigational risk by considering the identified navigational RIFs,the designed risk matrix,the navigational experiments of a MASS model,as well as experts’ experience,and verify the constructed BBN model and quantify the navigational risk based on objective and subjective data.The main contents of this research are summarized as follows.(1)Navigational RIF identificationA 4P4F model is built for identifying MASS navigational RIFs.Based on this model and literature review,RIFs are listed with respect to a remote control MASS.According to experts’ judgement,4 types of navigational RIFs are distinguished and assigned into 4 kinds of navigational phases.(2)Navigational risk comparisonAn approach is developed to compare the navigational risk among three navigational control models.Firstly,three general accident scenarios are established,in which failure modes are connected sequentially or in parallel.Given a built scenario,the probability,the severity,and the capability of not detecting of each failure mode involved are evaluated within three navigational control models by experts using integer number from 1 to 10.Such evaluations are transformed into interval numbers as the input parameters of the proposed approach.The obtained results not only show the order of risks in these compared navigational control models,i.e.,manual control,remote control,and autonomous control,but also present their uncertainty.(3)Risk matrix designA framework is proposed to design a risk matrix generally.In the context of a remote control MASS,the potential risk influences within different risk levels are obtained using Interval Type-2 Fuzzy Analytic Hierarchy Process(IT2FAHP).According to a rule of discretion,the plotted Continuous Probability Consequence Diagram(CPCD)is converted into a discrete risk matrix,providing risk criteria for risk assessment on a remote control MASS navigation.(4)Navigational risk assessmentA BBN model is constructed,in the qualitative part,based on the results of navigational RIF identification,the observation in the MASS model navigational experiments,and experts’ elicitation.In the quantitative part,a method for calculating Conditional Probability Tables(CPT)is proposed.Based on designed risk matrix or experts’ judgement,CPTs of nodes in the BBN model are generated.Based on the MASS model navigational experiments or experts’ experience,the prior probabilities of nodes in the BBN model are obtained.Next,based on results calculated from the BBN model and records in the MASS model navigational experiments,the validity of the BBN model is verified by analyzing the sensitivity,and assessing the agreement and discrimination.Additionally,based on the output of the BBN model,an approach is proposed and applied to evaluate the effectiveness of simple and compound risk control options.Based on the above mentioned work,the innovations of this study are underpinned by three points.Firstly,to identify navigational RIFs,a 4P4F model is established from the view of 4 kinds of navigational phases and 4 types of risk influencing factors.This model not only indicates human factors are not eradicated completely towards the development of autonomous ship,but also reflects that the technology plays a significant role as a hazard source for the navigation of smart ships.Moreover,some RIFs are further distinguished as specific RIFs for a remote control MASS(without crews onboard)rather than conventional ship or fully autonomous ship.Secondly,to measure the navigational risk level,an approach to design a risk matrix is proposed.In the approach,the potential risk influences are quantified within different risk levels.The risk matrix designed for a remote control MASS highlights that in the high risk level,to prevent an adverse event to occur is more important than to undermine the consequence of such event.Lastly,to quantify the navigational risk,a BBN model is established with nodes from the environment,ship,ashore,communication,and the others.Combing the experts’ experience and MASS model trails,the navigational risk is quantified.The established BBN model is verified from sensitivity,consistence,and discrepancy.Additionally,based on the BBN model,an approach is proposed to evaluate the risk control options,improving the effectiveness of risk control options.